Replication Instructions

This replication archive includes all data and code necessary to replicate the published analysis, however the intermediate outputs are excluded due to the large size of the model outputs, which total about 10GB.  To fully replicate the analysis in the paper requires a computer with at least 16GB of memory, at least 10GB of free storage, and about two weeks of computer time.  

1) Install the "eliot" package from the source files provided.  This package provides the CARIRT() function, and depends on the package Rcpp, which in turn requires a working C++ compiler on your computer.  See https://cran.r-project.org/web/packages/Rcpp/README.html

2) Run Script 1.  This estimates the case specific preferences and saves them as a 3.5GB .Rdata object in Output/.

3) Run Script 2.  The creates plots of the case specific preferences and processes them for use in the models for authorship and joining.  The saved model objects from these models are about 6.5GB in total.  The figures presented in the paper are generated by this script.

To reduce the storage and time requirements, the number of MCMC iterations specified in the two script files can be reduced.  Shorter simulations give a good approximation of the results presented in the paper.